-
-
Check out our applications -
subscribe to RSS
Get email updates every time we post!
DOA – just swap the faulty ones a build in cabinets
http://www.server-guy.com/2007/07/06/reliability-of-hp-servers/
I expected there would be some minor issues to work through as there are with most new hardware these days. What I didn’t expect was for three of the six servers purchased to be D.O.A… That’s right, half of the new servers purchased had failed hardware out of the box.
Check this post out, it’s mentioning deploying HP blades and how several were dead on arrival.
Let’s not get too worked up, it’s an industry wide issue, I’ve had the problem with the HP and IBM blades, both were similar, and it really wasn’t an issue and I’ll tell you why. Let’s say I had 10 cabinets (48 HP blades in a cabinet, 42 IBM blades in a cabinet), I always accepted that there was going to be some blades that had:
- Faulty memory
- Faulty network cards (could be that the switch port on the switch needed reset or the patching was broken)
- Backplane/enclosure issues – the blade worked but it didn’t understand which slot/enclosure it was in etc
It’s easier to take the following approach. Write off cabinet 10, remove the faulty blades as you go, you really are better, building the blades that you can in bulk and leaving one cabinet/rack with the range of blades that wont power up or have hardware faults.
Consider that you might run diagnostics on every rack server you buy, when you start buying a blade based grid solution of so many blades, are you really wanting to run individually the diagnostics on the blades? Surely the build process will iron out the hardware issues, if it doesn’t build you know there’s a problem.
No related posts.
Related posts brought to you by Yet Another Related Posts Plugin.
Leave a Reply
Recent Posts
Categories
Recent Comments
- Data centers on Joining the cloud computing space
- Tom on BuildMyVirtual - our Self Service Virtualization Portal goes online for free download shortly
- scott Paradis on What's in a name - quite a lot you know
- martin237 on How to reboot the HP ILO/kick off a hung session
- Steve Khan on What's the ILO and how do I connect to it?
- Mahmoud Moataz on How to reboot the HP ILO/kick off a hung session
- ST2 on Virtual mobile handset please
- Allan on How to reboot the HP ILO/kick off a hung session
- Chris Rugh on Vendor engagement and real world opportunities
- Ray McLemore on HP Converged Infrastructure enhances trade capacity and execution
2 Comments
I’m the guy who wrote the article you’re commenting on here, and I felt compelled to reply to your post.
While I completely agree with what you are saying in when dealing with a truly large environment, building the blades in bulk and dealing with the hardware faults later, the environment I work in is much smaller. The six servers I discussed represent the physical expansion of my server environment for the remainder of the year (there will be many more virtual machines created, but no additional hardware). The 10 empty slots in the new blade chassis will likely not be filled for another 12-18 months… so as you may imagine having three of the six servers come in dead caused serious concerns.
Our previous HP blades (64 p-class blades) all except one came in functioning correctly, no failures out of the box. That was the quality I was expecting with the new class of HP blades. I was disappointed when HP didn’t live up to my previous experience.
DOA blades remain a problem that needs to be addressed by the different vendors, it’s most frustrating, and I’d heard that the C class blades are meant to be a lot better than the P class ones. It’s an ongoing thing affecting the industry as a whole, thanks for your comments!